Toward Zeitgeist-Aware Multimodal (ZAM) Datasets of Pro-Eating Disorder Short-Form Videos

Reliable identification of pro-eating disorder (pro-ED) content online suffers from two pervasive problems: (1) existing methods predominantly rely on text-based signals, failing to capture the inherently multimodal nature of multimedia content; and (2) these methods struggle to keep pace with the rapid evolution of references, memes, terminology, and contextual cues that underlie this content. Together, these limitations point to a gap: the absence of expert-annotated reference standards capable of supporting real-time research and robust multimodal detection model training for pro-ED content on short-form video (SFV) platforms.

To address this, we propose the development of zeitgeist-aware multimodal (ZAM) datasets, which are continuously curated collections of annotated multimodal pro-ED content with inclusion criteria that evolve alongside the memetic zeitgeist: the variable essence of what is considered pro-ED as new media and references come into the cultural zeitgeist and are absorbed and interpreted in online spaces.

We present a rationale for such datasets, outline approaches for their curation, and describe our progress toward that end.

This ZAM curation method may benefit stakeholders across several fields who are interested in how pro-ED sentiment is encoded and transmitted through SFV content across time, including for the purpose of responsive moderation efforts.